Better support for long running tests #630
Description
This is somewhat similar to #169.
Long running tests could do with an ETA estimate of how long they are going to take. This is particularly the case when running with large --fuzz values, as the test time is relatively predictable (anecdote: running my test suite with --fuzz 10000 took 11,462ms; running it with --fuzz 1000000 took 1,214,796 ms, so a 100x increase in fuzz resulted in 105x increase in time, so this is pretty much linear).
The other thing is relatively minor, but showing elapsed time in ms is pretty... er... optimistic. Also difficult to read - Duration: 1214796 ms. Much easier would be Duration: 20 m 14 s.